Vitamin D Receptor Gene Polymorphisms Are Associated with Abdominal Visceral Adipose Tissue Volume and Serum Adipokine Concentrations but Not with Body Mass Index or Waist Circumference in African Americans: The Jackson Heart Study^,,

Abstract

Background: The biological actions of vitamin D are mediated through the vitamin D receptor (VDR). Single-nucleotide polymorphisms (SNPs) in the VDR gene have been previously associated with adiposity traits. However, to our knowledge, few studies have included direct measures of adiposity and adipokine concentrations.

Objective: We examined the association of tagging SNPs in the VDR gene with multiple adiposity measures, including waist circumference (WC), body mass index (BMI), body fat percentage, subcutaneous and visceral adipose tissue (VAT) volume, and serum adipokine (adiponectin and leptin) concentrations in adult African Americans (AAs).

Methods: Data from 3020 participants (61.9% women; mean age, 54.6 y) from the Jackson Heart Study were used for this analysis. Forty-five tag SNPs were chosen with the use of genotype data from the International HapMap project. We used linear regression to test the associations of imputed VDR SNPs with each of the traits, adjusted for age, sex, educational status, physical activity, smoking, alcohol intake, serum vitamin D concentration, European ancestry, and multiple testing.

Results: The G allele of the SNP rs4328262 remained associated with increased VAT volume after multiple testing correction (β = 45.7; P < 0.001). The A allele of another SNP (rs11574070) was nominally associated with body fat percentage (β = 0.96; P = 0.002). None of the VDR SNPs analyzed showed any link with WC or BMI. The A allele of rs2228570 (β = 0.08; P = 0.001) for men and the T allele of rs2853563 (β = 0.04; P < 0.001) for women remained positively associated with serum adiponectin concentrations after multiple testing correction.

Conclusion: Although we did not find any association for anthropometric measures, we did observe associations of VDR variants with serum adipokines and with the more metabolically active fat, VAT. Therefore, our findings demonstrate a possible role of VDR variants in regulating adipose tissue activity and adiposity among AAs.

Introduction

African Americans (AAs)⁶ have the highest rates of obesity and vitamin D deficiency in the United States (1). Vitamin D or serum 25-hydroxyvitamin D concentration is correlated inversely with adiposity and metabolic disorders (2–4). Based on the disproportionately high rates of vitamin D insufficiency and obesity among AAs, one potential candidate gene that may contribute to these disparities is the vitamin D receptor (VDR) gene. According to current evidence, vitamin D receptor is one of the most important key mediators for the biological activities of vitamin D; the interaction of vitamin D and this receptor plays an important role in adipocyte differentiation both in vitro and in vivo (5, 6). The active form of vitamin D (1,25-dihydroxyvitmain D) binds with the nuclear or cytosolic vitamin D receptor, which is located on chromosome 12 and contains 14 exons (chr12q13 · 1). This vitamin D-receptor compound functions as a ligand-activated gene transcription factor and can thus modulate the transcription of target genes (7, 8).

Several polymorphisms in the VDR gene that produce variations in the activity of vitamin D have been described in numerous populations (9–17). Studies to date have demonstrated that some VDR single-nucleotide polymorphisms (SNPs) such as rs1544410 (BsmI), rs7975232 (ApaI), rs731236 (TaqI), rs2239179, and rs3782905 are associated with different adiposity phenotypes (9, 15, 17–22). Most of these studies reported associations with indicators such as waist circumference (WC) or BMI but did not account for more precise measurements of regional fat accumulation such as subcutaneously stored abdominal fat [i.e., subcutaneous adipose tissue (SAT)] or adipose tissue located in the abdominal cavity and retroperitoneal area, which is commonly known as intra-abdominal or visceral adipose tissue (VAT). To our knowledge, none of these studies has examined VDR gene polymorphisms in relation to adipokine concentrations.

Although it has been established that vitamin D deficiency is more prevalent among AAs and the synthesis and action of vitamin D are influenced by the genetic background or SNPs in the VDR gene, to our knowledge not much has been explored about the relation between VDR gene polymorphisms and adiposity in this ethnic group. We speculate that investigating these possible associations in the AA population can help to identify potential prognostic variables and may predict an increased risk of developing adiposity in this population. This study aimed to examine the associations of tagging SNPs in the VDR gene with multiple adiposity traits such as BMI, WC, serum adipokine concentrations (adiponectin and leptin), body fat percentage, and body fat distribution (VAT and SAT) in adult AAs.

Methods

Data source.

Data for this analysis were obtained from the clinic visit during exams 1 and 2 of the Jackson Heart Study (JHS). JHS is a large community-based observational study initiated in 2000 to prospectively investigate the epidemiology and determinants of cardiovascular diseases in AAs (23). JHS comprises 5301 AA participants aged 21–94 y who were recruited between September 2000 and March 2004 from the Jackson, Mississippi, metropolitan area. Approximately 31% of JHS participants were recruited from the ARIC (Atherosclerosis Risk in Communities) study. Thus, they were common participants of JHS-ARIC. Of the total 5301 participants, 3029 or 57.1% (892 from the combined JHS-ARIC and 2136 specific to JHS) consented to genetic analyses and were genotyped separately from their parent studies in the CARe (Candidate Gene Association Resource) consortium with the use of the Affymetrix 6.0 platform (24). Of 3029 participants, 3020 had information on BMI, WC, body fat percentage, and adipokine concentrations. As part of exam 2, a subset of 1661 participants underwent multidetector computed tomography (CT) scanning from 2007 to 2009 and had information on body fat distribution (VAT and SAT). JHS was approved by the University of Mississippi Medical Center Institutional Review Board, and the participants gave written informed consent. Details of the study design and data collection methods have been described elsewhere (25, 26).

Measurement of adiposity traits.

Information about participants’ WC (cm), BMI (kg/m²), serum adiponectin (ng/mL), serum leptin (ng/dL), and body fat percentage was collected from the exam 1 data. Two measures of the waist at the level of the umbilicus, in the upright position, were averaged to measure WC for each participant. Weight was measured on a balance scale, in light clothing, without shoes, and recorded to the nearest 0.5 kg. Adiponectin and leptin concentrations were assayed with the use of standard protocols from the University of Minnesota (27). Body fat percentage was assessed with the use of a Tanita TNF 300A impedance body composition analyzer. Information on VAT and SAT volume was obtained from multidetector CT measurements for visceral adiposity, which was done during exam 2. The CT protocol for VAT and SAT measurements has been reported previously (28). Briefly, the CT images included a series through the lower abdomen detected by a Lightspeed 16 Pro CT system equipped with cardiac gating (GE Healthcare). The acquired abdominal imaging slices covering the lower abdomen from L3 to S1 were used for assessing VAT and SAT volume.

Assessment of covariates.

Information on participants’ age, sex, educational status, physical activity, smoking status, alcohol consumption, total calorie intake, and serum vitamin D concentration was obtained from exam 1 data. Educational status was self-reported and divided into 3 categories: less than high school, high school/some college, and college/associate degree or higher). The physical activity index composite score was calculated as the sum of 4 different domains of physical activity: active living, work, home and garden, and sport and exercise (29). A total score was the sum of these domains, with a maximum score of 24. A higher score indicated a higher level of total physical activity. Smoking status was classified as never, current, or former. Alcohol consumption status was defined as “yes” if the participant currently consumed alcoholic beverages and “no” if they had stopped drinking for more than a year or had never consumed alcohol. Serum concentrations of 25-hydroxyvitamin D₂ and 25-hydroxyvitamin D₃ were quantified with LC-tandem MS and were analyzed with isotope dilution-tandem MS (Quattro Micro with a 2795 HPLC; Waters). Serum concentrations were combined to obtain the serum total 25-hydroxyvitamin D. Global-proportion European ancestry estimates for our sample were calculated with the use of HAPMIX (v1.1) in analyses supported by the CARe consortium (30, 31). HAPMIX was run in a mode that assigns European or African ancestry to each allele, thus resolving the local ancestry of each allele when both genotype and local ancestry were heterozygous.

Tag SNP selection.

Potentially functional tag SNPs for VDR genes were chosen with the use of genotype data from the International HapMap collected in individuals of Yoruba ancestry (HapMap data phase 2 release 24, National Center for Biotechnology Information B36 assembly, dbSNP b126) (32). Haploview software version 3.3 was used to assess the linkage disequilibrium structure between SNPs (33). Tagger software was used for pairwise SNP selection with a minor allele frequency (MAF) >0.05 in individuals of Yoruba ancestry. It captured 100% of alleles with an r² of ≥0.8. In the tag SNP selection, we included previously studied functional SNPs of the VDR gene (rs731236, rs1544410, rs7975232, rs2239179, and rs3782905) before running tagger (9, 10, 15, 17). We also added 1 candidate SNP (rs2228570; FokI) from existing literature (11–13). This process resulted in a selection of 65 tagging SNPs for the VDR gene. For genotype imputation, IMPUTE2 software and reference-phased data from the 1000 Genomes Project Consortium (2012) were used (34). Details regarding the generation of the data can be found in the phase 1 article (35). Related individuals were removed before genotype imputation. An entire SNP imputation was done to combine and harmonize data content from the common JHS-ARIC-study genotyping platforms with the JHS-specific genotyping platform so that all data could be analyzed together (36). A more detailed quality control description of genotyping and imputation methods is presented in Supplemental Methods and Supplemental Table 1.

Statistical analysis.

Descriptive statistics were used for summarizing characteristics of the participants and were compared with the use of a chi-square or independent t test. We conducted a χ2 goodness of fit test to assess whether the studied SNPs were in Hardy-Weinberg equilibrium with the use of thresholds of P > 0.001. SNPs were removed from analyses if there was a deviation from the Hardy-Weinberg equilibrium or if they had an MAF <5%. Twenty SNPs were thus excluded, leaving 45 SNPs for the final analysis. We tested for associations of multiple adiposity measures with imputed VDR SNPs with the use of an additive genetic model. Natural logarithm was used to transform slightly skewed variables (adiponectin and leptin) for an improved approximation of normal distribution. The genetic associations with the continuous outcomes (BMI, WC, body fat percentage, VAT, SAT, adiponectin, and leptin) were examined with the use of 7 separate linear regression models, adjusted for age, sex, smoking status, alcohol use, physical activity, educational status, and European ancestral proportion estimates to account for population stratification. Models for BMI and WC were adjusted for total calorie intake as well. Regression models with body fat percentage, VAT, SAT, adiponectin, and leptin were additionally adjusted for BMI. A Bonferroni correction was used to adjust for multiple testing, with a corrected P value of 0.05 or 0.001 considered statistically significant for the 45 SNPs tested. Because of the differing body composition in men and women, we also tested for the interaction between the tag SNPs and sex on the adiposity traits. To evaluate the interactions, we added main effects and cross-product terms to the regression after adjusting for the same covariates mentioned previously. F tests comparing full and reduced models (with and without the interaction term) were performed to test the statistical significance of the interaction terms.

Results

The study sample included 3020 individuals, of which 1152 (36.5%) were men with a mean ± SD age of 54.0 ± 13.0 y, and 1868 (63.5%) were women with a mean age of 55.1 ± 12.7 y. Sex-specific characteristics of the participants are presented in Table 1. Educational status did not differ between men and women. Approximately 40% of the participants from both sexes had a college education or beyond (P = 0.79). A clear difference in sex, however, was found in alcohol use and smoking, with women being far less likely than men to consume alcohol and smoke cigarettes (P < 0.001 for all). Men were more physically active than women. Daily caloric intake and serum vitamin D concentrations were lower among women than men (P < 0.001 for all). Women also had a higher BMI, serum adiponectin, serum leptin, and body fat percentage than men (P < 0.001 for all). Table 1 also shows the compartmental fat distribution by sex (n = 1662; 599 men and 1063 women), in which women had a higher volume of SAT but lower volume of VAT than men.

TABLE 1.

Characteristics of African American participants by sex¹

	Men (n = 1152)	Women (n = 1868)	P2
Age, y	53.9 ± 13.0	55.1 ± 12.7	<0.05
Educational status, %
<High school	16.7	16.6
High school/some college	43.2	44.4
≥College/associate degree	40.1	39.0
Smoking status, %
Never	56.2	73.2
Former	25.7	15.8
Current	18.0	11.0	<0.001
Alcohol consumption, %
No	14.0	32.9
Sometimes	26.8	26.8	<0.001
Yes	59.2	40.3
Total physical activity score3	8.55 ± 2.56	8.15 ± 2.57	<0.001
Total energy intake, kcal/d	2519 ± 1353	1985 ± 1124	<0.001
Global proportion of European ancestry	0.18 ± 0.09	0.16 ± 0.08	<0.05
Serum vitamin D, ng/mL	15.3 ± 6.5	13.6 ± 6.5	<0.001
BMI, kg/m2	30.0 ± 6.3	33.3 ± 7.8	<0.001
Waist circumference, cm	101.7 ± 5.4	101.1 ± 16.9
Serum adiponectin, ng/mL	4077 ± 3380	6037 ± 4458	<0.001
Serum leptin, ng/mL	11.8 ± 11.1	38.5 ± 24.8	<0.001
Body fat, %	29.6 ± 7.9	42.3 ± 7.0	<0.001
Visceral adipose tissue volume,4 mm3/50	879 ± 407	819 ± 365	<0.01
Subcutaneous adipose tissue volume,4 mm3/50	1727 ± 797	2685 ± 962	<0.001

¹Data are means ± SDs of subjects unless otherwise indicated, n = 3020.

²Independent t test or chi-square test.

³Sum of the 4 different domains of physical activity: active living, work, home and garden, and sports and exercise indexes.

⁴For 50 mm of abdomen.

The MAFs, physical positions, imputation quality of the 45 VDR SNPs, and the covariate-adjusted associations of VDR with BMI and WC are presented in Supplemental Table 2. None of the tag SNPs showed significant associations with BMI or WC in the studied population. We found no evidence of effect modification by sex affecting associations between the tag SNPs and BMI or WC. There were also no statistically significant interactions between the tag SNPs and sex on body fat percentage, VAT, or SAT; the associations between all 45 VDR SNPs and these traits are presented in Supplemental Table 3. In our analysis, the G allele of SNP rs4328262 (MAF = 0.35) remained significantly associated with increased VAT volume after multiple testing correction (Table 2). For this SNP, assuming an additive genetic model on inheritance, we observed that each additional rare allele resulted in a ∼46-mm³ increase in VAT volume [β ± SE = 45.7 ± 15.2; nominal P value (not adjusted for multiple testing) or P = 0.001 and corrected P value (adjusted for multiple testing with the use of a Bonferroni correction) or P* < 0.05]. The A allele of another SNP, rs11574070, was marginally associated with body fat percentage (β ± SE = 0.96 ± 0.33; P = 0.002; P* < 0.09). None of the tag SNPs showed any link with SAT.

TABLE 2.

Linear regression-derived associations of selected tagging VDR SNPs with body fat percentage and VAT and SAT volumes among African Americans¹

	Body fat percentage (n = 3020)		VAT,2 mm3/50 (n = 1661)		SAT,2 mm3/50 (n = 1661)
SNPs	β ± SE	P (P*)3	β ± SE	P (P*)3	β ± SE	P3
rs11574070	−0.81 ± 0.29	0.002 (0.09)	−23.4 ± 20.8	0.26	−74.5 ± 35.9	0.03
rs4328262	−0.42 ± 0.22	0.06	45.7 ± 15.3	0.001 (<0.05)	−11.3 ± 26.4

¹Results are presented as β ± SEs, n = 3020; β coefficients represent the mean difference in body fat percentage, VAT, or SAT for each additional rare allele of the corresponding SNP. Three separate linear regression models, including each SNP alone, were adjusted for age, sex, BMI, smoking status, alcohol use, physical activity, educational status, and European ancestral proportion. SAT, subcutaneous adipose tissue; SNP, single-nucleotide polymorphism; VAT, visceral adipose tissue; VDR, vitamin D receptor.

²For 50 mm of abdomen.

³Nominal P values (not adjusted for multiple testing) and corrected P values (P*) (adjusted for multiple testing with the Bonferroni correction where significant, P < 0.10).

Based on an observed significant VDR gene-sex interaction on adiponectin and leptin concentrations, the sex-stratified associations of selected VDR SNPs with log-transformed adiponectin and leptin concentrations are presented in Tables 3 and 4 (associations with all 45 VDR SNPs are presented in Supplemental Table 4). We identified a statistically significant positive association for rs2228570 (MAF = 0.24) in men and for rs2853563 (MAF = 0.17) in women with log-transformed adiponectin concentrations. For SNP rs2228570 (β ± SE = 0.08 ± 0.03; P = 0.001; P* < 0.05), we observed that each additional rare allele (A allele) resulted in an 8% increase of adiponectin in men, and for SNP rs2853563 (β ± SE = 0.04 ± 0.0; P < 0.001; P* < 0.01), each additional rare allele (T allele) resulted in a 4% increase of adiponectin in women. None of the tag SNPs showed any significant associations with leptin concentrations either for men or women after correcting for multiple testing. Nominal P values were significant, however, for the SNPs rs7975232, rs2525044, rs7975128, rs7305032, rs11168266, rs11168268, and rs987849 for men and SNPs rs11574138, rs11574070, rs12717991, rs4760648, and rs4334089 for women.

TABLE 3.

Sex-stratified association of selected tagging VDR SNPs with log-transformed serum adiponectin concentrations among African Americans¹

	Men (n = 1152)		Women (n = 1868)
SNPs	β ± SE	P (P*)2	β ± SE	P (P*)2
rs2853563	−0.015 ± 0.01		0.044 ± 0.01	<0.001 (<0.01)
rs7967673	−0.029 ± 0.02		0.042 ± 0.01	0.007
rs11168266	0.007 ± 0.01		0.018 ± 0.01	0.05
rs11574070	−0.043 ± 0.02	0.05	0.029 ± 0.01	0.06
rs2228570	0.08 ± 0.03	0.001 (<0.05)	0.011 ± 0.02
rs4328262	−0.005 ± 0.01		−0.028 ± 0.01	0.04

¹Results are presented as β ± SEs, n = 3020; β coefficients represent the mean percentage difference in adiponectin concentrations for each additional rare allele of the corresponding SNP. Two separate linear regression models, including each SNP alone, were adjusted for age, sex, BMI, smoking status, alcohol use, physical activity, educational status, and European ancestral proportion. SNP, single-nucleotide polymorphism; VDR, vitamin D receptor.

²Nominal P values (not adjusted for multiple testing) and corrected P values (P*) (adjusted for multiple testing with a Bonferroni correction where significant, P < 0.05).

TABLE 4.

Sex-stratified association of selected tagging VDR SNPs with log-transformed serum leptin concentrations among African Americans¹

	Men (n = 1152)		Women (n = 1868)
SNPs	β ± SE	P2	β ± SE	P2
rs11574138	0.014 ± 0.021		0.03 ± 0.01	0.04
rs7975232	0.023 ± 0.01	0.03	0.008 ± 0.01
rs2525044	−0.024 ± 0.01	0.05	−0.006 ± 0.01
rs7975128	0.028 ± 0.012	0.04	0.01 ± 0.01
rs7305032	−0.023 ± 0.01	0.05	−0.12 ± 0.01
rs11168266	−0.024 ± 0.01	0.02	−0.006 ± 0.01
rs11168268	−0.021 ± 0.01	0.04	−0.006 ± 0.01
rs987849	−0.029 ± 0.01	0.02	−0.002 ± 0.01
rs11574070	0.017 ± 0.02		−0.02 ± 0.01	0.05
rs12717991	0.012 ± 0.01		−0.016 ± 0.01	0.03
rs4760648	0.014 ± 0.01		−0.018 ± 0.01	0.04
rs4334089	0.014 ± 0.01		−0.015 ± 0.01	0.05

¹Results are presented as β ± SEs, n = 3020; β coefficients represent the mean percentage difference in leptin concentrations for each additional rare allele of the corresponding SNP. Two separate linear regression models, including each SNP alone, were adjusted for age, sex, BMI, smoking status, alcohol use, physical activity, educational status, and European ancestral proportion. SNP, single-nucleotide polymorphism; VDR, vitamin D receptor.

²Nominal P values (not adjusted for multiple testing).

Discussion

We found no significant link between VDR SNPs and anthropometric measures (WC and BMI), but we identified a positive association of one VDR SNP (rs4328262) with VAT, which survived the adjustment for multiple testing. Furthermore, another SNP, rs11574070, had a borderline positive association with body fat percentage. Two of the SNPs, rs2853563 for women and rs2228570 for men, respectively, were also positively associated with adiponectin concentrations after multiple testing correction. Hence, our study provides evidence that polymorphisms in the VDR gene might play a role in regulating adipose tissue activity body fatness and susceptibility to adiposity among AAs. Studying VDR genetic variants for associations with adiposity measures in the AA population is important given the overall higher obesity prevalence as well as lower vitamin D concentrations in AAs than in other US populations.

Genetic variants in the VDR gene have been previously associated with adiposity-related phenotypes (9, 15, 17–22). However, almost all of the studies, to our knowledge, were performed primarily on Caucasian or Asian cohorts and did not report associations in people of African descent. Studies from Sweden, France, Poland, China, and Saudi Arabia identified SNPs rs1544410, rs17878969 [poly(A)], rs7975232, and rs731236 to be associated with BMI and WC (9, 15, 18). In an analysis from the United States, Ochs-Balcom et al. (17) detected considerable differences in adiposity phenotypes by genotype for the VDR SNP rs3782905 among a sample of 1773 healthy Caucasian women. With a smaller sample of ∼250 participants, another recent study from the United States (20) reported significant associations of SNPs rs11568820 and rs7975232 with central adiposity among Caucasian women. On the other hand, a study involving 6922 Chinese women and another among 5224 individuals from the 1958 British birth cohort found no evidence of any associations between VDR SNPs and adiposity-related traits (21, 22). Many of the studies discussed previously only considered polymorphisms of convenience, such as restriction fragment-length polymorphisms, including rs731236, rs1544410, and rs7975232, or focused on possible functional variants such as rs2228570 (FokI) or rs11568820. Only a few studies, to our knowledge, have tested SNPs that tag most of the common genetic variations in VDR with the use of a systematic tag SNP approach (17, 22). Almost all of these studies used anthropometric measures such as BMI or WC as obesity measurements. A limitation of these anthropometric measures is that they do not distinguish fat from lean mass and are unable to characterize the type and distribution of fat deposits. Because VATs are a metabolically active fat depot, this is more closely tied to the metabolic consequences of adiposity than other measures (37, 38). The use of CT scanning allowed us to quantify VAT directly and precisely, which is a more refined and objective indicator of adiposity. Thus, although we did not detect any associations for anthropometric measures, our observed association of VDR SNP rs4328262 with VAT supports the notion that the VDR gene is likely to be related to the development of obesity and obesity-related outcomes. To our knowledge, this observed association is novel. We also detected a borderline association of the SNP rs11574070 with body fat percentage. Both of these SNPs are located in the intronic region at the 3′ end of the VDR gene. Although introns are noncoding regions, intronic SNPs can potentially affect protein synthesis by influencing RNA splicing. Moreover, introns may have their own transcription units that produce regulatory RNAs or perhaps small proteins, in which case SNPs may affect the functions of the corresponding products of intron expression (39). Polymorphisms in the 3′ untranslated region (UTR) regulate VDR gene expression by modulating messenger RNA stability (40). Thus, the intronic SNPs may still alter the function of the VDR gene if the SNPs are in high-linkage disequilibrium with an allele of a truly functional sequence variation in the 3′ UTR region (40, 41). Thus, the association of VDR SNP with VAT in our study suggests that the vitamin D receptor is possibly involved in developing adiposity in AAs. Multiple mechanisms have been suggested through which an active form of vitamin D in combination with their receptors can potentially play a role in the etiology of adiposity. One of the main suggested mechanisms is that vitamin D modulates adipogenesis through vitamin D receptor-dependent molecular components of adipogenesis, such as PPAR-γ and enhancer-binding protein α, and thus inhibits corresponding adipocyte differentiation (42). In transgenic mice studies, overexpressed human vitamin D receptor in adipocytes led to marked decreases in energy expenditure and the induction of obesity (5). In in vitro experiments, vitamin D also inhibited the early phase of differentiation of preadipocytes to mature adipocytes by interacting with this receptor (6).

In addition to visceral fat, we also detected a significant sex-specific association of the VDR polymorphism with serum adiponectin concentrations. Adiponectin increases FA uptake, slows FA synthesis, decreases gluconeogenesis, and is inversely related to obesity (43, 44). We observed the T allele of rs2853563 for women and the A allele of rs2228570 for men to be associated with increased adiponectin concentrations. Among these polymorphisms, rs2228570 is found in the promoter region and defines 2 potential translation initiations that result in a shorter protein with increased biological activity (45). SNP rs2853563, on the other hand, is located in the 3′ UTR and may influence the binding of vitamin D and the vitamin D receptor by modulating messenger RNA stability (40). rs2228570 is a well-studied variant and has mainly been associated with different forms of cancer (14, 46). Regarding cardiovascular risk factors, this variant has mostly been associated with hypertension (47, 48), with some small studies indicating its impact on insulin sensitivity, blood lipid concentrations, and BMI (12, 13, 15). However, to our knowledge, neither rs2228570 nor rs2853563 has previously been shown to be associated with serum adiponectin concentration. As mentioned previously, in this analysis we detected a sex-specific association of these 2 VDR polymorphisms with adiponectin concentration. The sexual dimorphism in adiponectin concentrations is well established and seems to depend mainly on gonadal steroids (49). This concept is further supported by findings in mice that suggest that gonadal steroid hormones play important roles in the sex-specific expression of genes related to metabolic traits (50). In addition, one of our previous studies conducted in the same population also reported a significant sex-specific association of ADIPOQ polymorphism with serum adiponectin concentrations (51).

Although most functions of vitamin D are mediated by the VDR gene, there are other genes that may contribute to the genetic basis of vitamin D status by controlling pathways that synthesize, transport, and degrade forms of vitamin D (52, 53). Some of the important genes involved in the vitamin D pathway include 7-dehydrocholesterol reductase; 25-hydroxyvitamin D-25 hydroxylase; vitamin D-binding protein; 25-hydroxyvitamin D-1a hydroxylase; cytochrome P450, family 27, subfamily A, polypeptide 1; 25-hydroxyvitamin D-24 hydroxylase; and retinoid X receptor γ (52–54). Without having genotype data of these pathway genes, we were unable to discern whether polymorphisms of these genes could also explain the reported association between vitamin D and adiposity. Because our study sample was limited to AAs, we acknowledge that the results may not be generalizable to other ethnic groups. The lack of a replication group to confirm our findings was another limitation of our study, and it will be important to pursue replication studies in other samples of AAs. Despite such limitations, our study is one of the first to our knowledge to evaluate the relation between an extensive number of VDR SNPs and different adiposity traits in AAs. For SNP selection, the tag approach was chosen, and multiple testing correction was applied with the use of a Bonferroni correction for the number of SNPs tested to maximize genetic coverage and to make the results reliable (55). One of the main strengths of our study was our ability to directly measure and partition the subcutaneous from visceral fat deposits with the use of CT imaging. Thus, we were able to show the association between the polymorphism and the most metabolically active form of fat, which cannot be captured by conventional anthropometric measurements. Furthermore, all of our multivariable analyses accounted for estimated individual ancestry, which enabled us to reduce type 1 error or false-positive results. This is important because genetic-association analyses in admixed populations such as AAs have highlighted the complexities and biases introduced in the analyses because of confounding by ancestry (56). Another major strength of our study was that our data from a community-based AA population was not selected on the basis of adiposity-related traits or vitamin D status, and we had a uniform technique to measure all social, physiological, and biochemical values.

Based on our observed associations between VDR variants and adiposity, we recommend that future studies investigate the relations between other vitamin D pathway genes and adiposity in AAs. This may indicate a potential need to screen for vitamin D-related gene polymorphisms in the context of obesity, particularly before considering vitamin D supplementation. Further research to identify the functionality of VDR gene polymorphisms and their potential interactions with sex may also improve our understanding of the underlying mechanisms as to how the vitamin D endocrine system is involved in adipocyte differentiation and metabolism. Investigating the vitamin D system with regard to racial disparity in adiposity-related outcomes would also be interesting.

In conclusion, our study suggests that polymorphisms in the VDR gene are associated with susceptibility to adiposity in AAs. The genetic factors that contribute to adiposity are certainly more complex than to be explained entirely by variations in a single gene. Nevertheless, our results, together with more genetic epidemiological studies in larger populations, will help to further understanding of the biological significance of vitamin D-related polymorphisms in relation to adiposity and could contribute to the identification of novel therapeutic strategies to prevent or treat adiposity and adiposity-related disorders among AAs.